Abstract

Although there has been much research on the application of program slicing to the problem of software testing, most of it has focussed on regression testing. The objective of the published techniques is to reduce its cost by identifying the set of existing test cases which are guaranteed to exercise the modified program components. In this research, program slicing is applied to behavioral testing. Three testing approaches are presented to ensure that modifications made to a program have not adversely affected its correct behavior. The proposed testing techniques, as well as the underlying dynamic program slicing algorithm, are implemented as part of the CONCEPT research project. A case study using the JUnit testing framework is also presented to demonstrate their applicability in detecting faults, which could escape from traditional testing techniques.

Document Downloads